Power supply build quality pictorial. part 2

[goodpsusearch]

Re: Power supply build quality pictorial. part 2

And the fan controller is better, fan rpm at idle way lower.

I wouldn't call it better.

[Behemot]

Re: Power supply build quality pictorial. part 2

Good that the rest of the planet would

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

I wouldn't call it better.

Because you like loud PSU?
There is absolutely no reason to let the fan rotate with ~1000rpm when the unit isn't loaded.

Or do you prefer no fan with lower loads like for example the new Cooler Master and many somewhat decent Corsair things?? And Seasonic started that shit...

As for load: no worrys, fan speed increases by 3 times or so under higher load (1350rpm or somethin like that)...

PS: the worst one I have here is the SF Leadex II wich had around 800rpm when the fan started...
And the Super Flower Platinum King - 900rpm but with an 120mm fan...

[goodpsusearch]

TurboX 650W (DR-8600BTX)

TurboX 650W (DR-8600BTX)

This is actually a decent 350W continuous 400W peal (maybe) power supply. Unfortunately the label claims it is an 650W unit.

The fan is Globe fan sleeve bearing. Not seized, probably because it is thermally controlled with an add on board mounted on the secondary heatsink.

It is a PPFC unit. The Passive PFC coil is not fake.

All the input filtering is on an extra board. It includes:
2 common mode chokes
4 Y caps
2 X caps
sleeved thermistor
fuse

There are also 2 varistors near the primary caps.

The primary capacitors are ANODIA 680uF, their real capacity was measured 650uF and 670uF. That's good.

The 4A bridge rectifier is more than enough for 230V mains.

The main transformer is real ERL 35 and the 5vsb transformer is 19 size.

The main switching transistors are 2x D209L, actually better than 13009 and they come in a nice TO-3P package.

I was surprised to see that the 5vsb uses a DL0165R IC and not the 2 switching flyback circuit.

The output is rectified with SB5A60F that is bolted on secondary heatsink. I couldn't find a datasheet, but it seems to be a 5A 60V diode. Nice!

PWM controller/supervisor combo IC is 2012 ATX chip of the year. No pdf available online

The soldering looks quite good. There was plastic insulation bellow the PCB.

3.3V and 5V use schottky rectifier SB3045ST.

12V uses SBL3060PT schottky rectifier with 0.7V voltage drop.

There are pi coils installed on every output rail but not -12V.

-12V rail caps:
470uF 16V yc (saturn)

5vsb: 2x1000uF 10V yc (saturn) LE

3.3V: 2x2200uF 10V yc (saturn) LE

5V: 2x2200uF 10V yc (saturn) LE

12V: 2x1000uF 16V yc (saturn) LE

The heatsinks are thick and have fins, I think they are good for 350W or 400W for some seconds.

I am going to do a partial recap, oil the fan and check the fan controller.

[momaka]

Re: TurboX 650W (DR-8600BTX)

TurboX 650W (DR-8600BTX)

This is actually a decent 350W continuous 400W peal (maybe) power supply. Unfortunately the label claims it is an 650W unit.

Ha! Yeah, this ain't no 650W unit FOR SURE!

Heck, I think even 350W continuous might a be a bit too much. Firstly, it's got a 35 mm core main transformer, so that's an automatic 300-350 Watt limitation for an old half-bridge with a slow switching frequency. The output filtering with those 1000 uF caps is also inadequate for anything more. Likewise, the primary caps being 680 uF is also probably a limitation to about 350 Watts before the hold-up time drops too low.

If that's not convincing, I acquired a very similar Allied from nearly the same time period ("2012" PWM chip, and almost identical output filtering stage). I got it because it is blown - both the primary BJTs and the fuse. From the information I remember getting about it, that PSU was powering a fairly new i5 or i7 system with an AMD R9 290 or similarly power-hungry GPU. My guess would be the 12V rail for that system was probably peaking at 300-350 Watts (which you can say was probably the power consumption for the entire system) - and yet the PSU blew. So I'll stick with my guess that this is a 300W PSU with *maybe* 350W peak in its "reserve".

But for a more basic PC, this would be plenty. The 12V rectifier was rated for 30 Amps IIRC, but I'll check. Heck, I have pictures of it. I'll see if I can post it later / in a few days.

The 4A bridge rectifier is more than enough for 230V mains.

I bet Deer/Allied used the same thing on the 110/120V versions, but I will check mine, of course. If they did, then you know it's likely a 300W unit at best.

The main switching transistors are 2x D209L, actually better than 13009 and they come in a nice TO-3P package.

Yup, that's what I think mine has too - D209L in TO-3P. Blown!

The soldering looks quite good.

I agree.
Deer/L&C/Solytech have gotten pretty good with their soldering. Shame they still make these crappy PSUs... or at least let them get over-rated so much. With good caps, these PSUs are solid. Maybe not very efficient and don't have the best of protections, but they keep on working.

I am going to do a partial recap, oil the fan and check the fan controller.

Ha! That's exactly how I usually do it with these lower-tier PSUs as well: partial recap with good quality Japanese caps - that way, there's at least one good cap in the output filter, if the other happens to fail. And I end up saving my PSU caps for better-built PSUs, like Delta, HiPro, FSP, and LiteON.

https://www.badcaps.net/forum/showthread.php?t=23996

They are not suitable for 12V powered machines

EDIT: Check this pdf
https://cdn.badcaps-static.com/pdfs/...a1ac9621a8.pdf

There are some pictures at the last pages.

Thanks for the info!

Yeah, I know they are not suitable for 12V systems indeed. Might take apart my Task PSU someday and play with its compensation and filtering to see if I can add that 12V filter cap and make the PSU stable. There has to be a way to fix that!

Something new, 80plus Gold (230VAC) certified, possibly the cheapest one there is:



Not bad for the price that is under a Corsair CX750M - almost 10€...
And the fan controller is better, fan rpm at idle way lower.
I've measured something around 400rpm or something like that. And ~1400rpm at ~700W Load.

Interesting PSU. Looks good, aside from the cheaper caps (though that is to be expected with such a cheap PSU, of course) - at least it can do it's labeled power. I wonder how long it will last, though. I still can't bring myself to trust these APFC PSUs - seems they always have problems after x amount of years... or at least a lot more often than the simple PFC-less PSUs.

[goodpsusearch]

Re: Power supply build quality pictorial. part 2

I tested the Turbo X power supply in a Pentium 4 3GHZ HT, 2 hdd drivers and Geforce 6600 PCI-E. I disconnected case fans and ran IBT + Furmark for more than an hour. The psu got warm but not hot at all. According to my online UPS I use to estimate watts of connected devices, it drew 210W.

Partial Recap
12V:
I replaced 1000uF caps with 1x 2200uF 16V Nichicon and 1x 3300uF 16V JunFu WG.

5vsb:
I also replaced 5vsb YC caps with better quality parts.

Fan mod
9.13kohm resistor in parallel with the thermistor
Psu remains silent without getting hot even when PC case cooling is intentionally compromised.

Minimum load resistors

Original --> New

3.3V: 15ohm (~0.75W) --> 27ohm taken from 5V (~0.43W)

5vsb: no minimum load resistor used

5V: 27ohm (~0.926W) --> 120ohm taken from 12V (~0.22W)

12V: 120ohm (~1.2W) --> 240ohm (~0.62W)

-12V: 390ohm (~0.37W) --> no change

Finally, I oiled the fan with original Singer sewing machine oil.

Pics attached

[Dan81]

Re: Power supply build quality pictorial. part 2

I tested the Turbo X power supply in a Pentium 4 3GHZ HT, 2 hdd drivers and Geforce 6600 PCI-E. I disconnected case fans and ran IBT + Furmark for more than an hour. The psu got warm but not hot at all. According to my online UPS I use to estimate watts of connected devices, it drew 210W.

Partial Recap
12V:
I replaced 1000uF caps with 1x 2200uF 16V Nichicon and 1x 3300uF 16V JunFu WG.

5vsb:
I also replaced 5vsb YC caps with better quality parts.

Fan mod
9.13kohm resistor in parallel with the thermistor
Psu remains silent without getting hot even when PC case cooling is intentionally compromised.

Minimum load resistors

Original --> New

3.3V: 15ohm (~0.75W) --> 27ohm taken from 5V (~0.43W)

5vsb: no minimum load resistor used

5V: 27ohm (~0.926W) --> 120ohm taken from 12V (~0.22W)

12V: 120ohm (~1.2W) --> 240ohm (~0.62W)

-12V: 390ohm (~0.37W) --> no change

Finally, I oiled the fan with original Singer sewing machine oil.

Pics attached

Might be worth replacing the secondary rectifiers as well. I replaced the 5v one with a 16A rectifier and the 3.3 with a 20A rectifier, as well as the 12v rectifier with a 20A 60v part. Now it's powering a 2.8GHz Northwood just fine, and the air that comes out of it isn't really hot. (just warm).

[Stefan Payne]

Re: TurboX 650W (DR-8600BTX)

Interesting PSU. Looks good, aside from the cheaper caps (though that is to be expected with such a cheap PSU, of course) - at least it can do it's labeled power. I wonder how long it will last, though. I still can't bring myself to trust these APFC PSUs - seems they always have problems after x amount of years... or at least a lot more often than the simple PFC-less PSUs.

Absolutely, was pleasantly surprised.

But what do you say about the polymer caps in +12V (those 8mm Types), are they Nichicon??
Minor rails are entirely filtered by Polymers, +12V only has two wet lyrics (2200uF/16V, like the ones you see in +5VSB).

Though there are some caps I don't get - like the light green ones wich are 220uF/35V types.



PFC is done by a single switcher and a single diode. Though it's a Full Bridge LLC Resonant design (4 MOSFETs for the primary).


Its really cramped but to me it seems like the engineers developing this unit were really good ones.
Though I fear that it is possible that +12V rectifiers might cook the +5VSB caps.

[goodpsusearch]

Re: Power supply build quality pictorial. part 2

Might be worth replacing the secondary rectifiers as well. I replaced the 5v one with a 16A rectifier and the 3.3 with a 20A rectifier, as well as the 12v rectifier with a 20A 60v part. Now it's powering a 2.8GHz Northwood just fine, and the air that comes out of it isn't really hot. (just warm).

All the secondary rectifiers are Schottky 30A with low forward voltage drop

[momaka]

Re: Power supply build quality pictorial. part 2

I tested the Turbo X power supply in a Pentium 4 3GHZ HT, 2 hdd drivers and Geforce 6600 PCI-E. I disconnected case fans and ran IBT + Furmark for more than an hour. The psu got warm but not hot at all. According to my online UPS I use to estimate watts of connected devices, it drew 210W.

Nice feature on that UPS.

Seems more or less accurate for that PC - that is, 210 *AC* Watts (i.e. from the wall plug). With an assumed 75% efficiency, this translates to a little under 160 Watts of DC power (and that more or less agrees with what's in that PC, as a P4 HT usually draws about 100 Watts under load, the 6600 draws another 30W, the HDD maybe 10W or less, the chipset another 10W or thereabouts, and the RAM usually around 5W too... so that comes up to around 155 Watts total, not counting fans and some other smaller stuff.)

Anyways, nice mods. Should be good enough to use as a basic PSU to test stuff around. Maybe even use it in a low-mid-power PC.

All the secondary rectifiers are Schottky 30A with low forward voltage drop

Yup, I don't see why goodpsusearch would replace them with 20 Amp ones.

Dan, I think you're too used to seeing 10A/15A rectifiers in those low-end L&C PSUs - they are skewing your expectations , man.

But what do you say about the polymer caps in +12V (those 8mm Types), are they Nichicon??

Oh nice, I didn't notice their info. I think Nichicon it is. But I'm not sure which series. IIRC, FP designates their "Functional Polymer" series. Not sure if they had more than one, though.

In any case, if those are genuine Nichicon / FP caps, that's pretty good.

*EDIT*
Yeah, those are Nichicon FP caps indeed. But which series of FP cap - don't ask me! I don't think there is a way to tell from the writing on the cap. So it could be FPCAP E5 series, or HT series, or NE series, or... (there's a few more besides these).

PFC is done by a single switcher and a single diode. Though it's a Full Bridge LLC Resonant design (4 MOSFETs for the primary).

That doesn't make me any more or any less at ease. I just feel that most APFC PSUs are too complicated for their own good. That's not to say that an APFC PSU can't/won't last long. But statistically speaking, they do have a higher chance of failure as there are more parts (to fail) in them.

Its really cramped but to me it seems like the engineers developing this unit were really good ones.

Only time will tell.

[goodpsusearch]

Re: Power supply build quality pictorial. part 2

Anyways, nice mods. Should be good enough to use as a basic PSU to test stuff around. Maybe even use it in a low-mid-power PC.

It came from an i5-3470 pc, Gigabyte Ultra durable 3 motherboard GA-B75M-D3H and 8GB DDR3, 1TB HDD and just 1 90mm fan. Oh, and almost forgot: Nvidia GT630 gpu

I am running some stress tests now (IBT and Furmark simultaneously) and even though the i5 dynamically overclocks to 3.6GHZ, the psu is not as warm as it were on the P4 system. According to UPS LCD it draws just 155W, 200VA.

I am returning it to customer today and getting paid

[Behemot]

Re: Power supply build quality pictorial. part 2

I don't bother with these craps any more. Even a decade old FSP unit which are still so very common around here are five times as robust even if the efficiency is similarly low.

With so many cheaper and better newer units with 10 % points better efficiency around, there is no point to refurbish that crap.

Cosindering the price to refurbish this and the extra electricity it takes, some new half-decent 350watter, 80 PLUS (Bronze) rated would pay itself in a year, most likely.

[goodpsusearch]

Re: Power supply build quality pictorial. part 2

65-75watt during normal use (no stress tests), what extra electricity are you talking about?

This is just marketing arguments from psu industry.

I got paid

[Behemot]

Re: Power supply build quality pictorial. part 2

About ten watthours, if it's running 8 hours a day, that already makes few bucks per year. I also expect the two hours you worked on it ain't free. More and more even cheap units have full sets of protections, plus 3y warranty becomes minimum even for low-end (not considering total garbage). There is night and day difference between half-decent forward unit and *this*.

OFC, it is better for you cause you have better margin on the work than on putting in a new PSU. The customer is who gets screwed in here…

[goodpsusearch]

Re: Power supply build quality pictorial. part 2

3 hours per day will be its use.

Judging from the fact that the psu remained cold during idle and normal use, I would guess 74-78% efficiency on 230V AC. Not too much of a difference from 80 plus.

And for that hours per day it doesn't matter at all.

Bad caps matter, but he would get crap caps anyway if he bought a new psu for 20-30€. Try convincing people here to spend more for a 2012 PC. He was thinking of recycling it!

[Behemot]

Re: Power supply build quality pictorial. part 2

Year for such short use that's likely OK. For an office machine that would make measurable difference though. You just hope they won't decide to give it to kids and upgrade to 100W+ graphics.

In most new units besides FSP with their custom PoS, caps do not matter that much anymore. It has been some time since I've seen more modern unit with more than maybe single bad cap (after what, 5-8 years of use?). The problem is still there, but reduced to maybe 10 % of what it used to be. When so many units have 5+ years of warranty, you do not have to consider caps that much anymore. If it fails, you just get a replacement.

IIRC, you recycled it for him, the only thing which remained was the case, the PSU and maybe a drive. With the price of the paper cases ppl buy, I understand why they throw away the whole rigs. Often the old generations have so poor cooling it really makes very little sense to keep it.

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

In most new units besides FSP with their custom PoS, caps do not matter that much anymore. It has been some time since I've seen more modern unit with more than maybe single bad cap (after what, 5-8 years of use?). The problem is still there, but reduced to maybe 10 % of what it used to be. When so many units have 5+ years of warranty, you do not have to consider caps that much anymore. If it fails, you just get a replacement.

Yeah and there are so much more things to go wrong - like the +5VSB Chip that seems to explode left and right sometimes...

[Stefan Payne]

Re: Power supply build quality pictorial. part 2

Have I posted these ones?
Controller seems to be a PIC - so some kind of Digitally controlled thing...

Have no idea how that thing works...
And it seems to regulate every voltage independently with its own transformer?!

[momaka]

Re: Power supply build quality pictorial. part 2

It came from an i5-3470 pc, Gigabyte Ultra durable 3 motherboard GA-B75M-D3H and 8GB DDR3, 1TB HDD and just 1 90mm fan. Oh, and almost forgot: Nvidia GT630 gpu

I am running some stress tests now (IBT and Furmark simultaneously) and even though the i5 dynamically overclocks to 3.6GHZ, the psu is not as warm as it were on the P4 system. According to UPS LCD it draws just 155W, 200VA.

Yeah, i5s and i7s are a lot more efficient than a P4. At idle, you can expect them to draw maybe 20-30 Watts, if even that. The P4 doesn't throttle too well, so it will still burn at the very minimum 50+ odd Watts and often idling around the 60-70 Watt mark. But then again, P4's are f#$%^&* indestructible!

I don't bother with these craps any more. Even a decade old FSP unit which are still so very common around here are five times as robust even if the efficiency is similarly low.

In general, I agree.

I myself prefer to refurbish units from OEM machines whenever possible (like Delta, HiPro, LiteON, and Bestec from Dell/HP/Gateway/Acer PCs.) But I wouldn't call these more "robust". I think a better way to describe them is to say they are safer (whereas the really cheap PSUs could be just as robust... with good caps and fan, of course..., but not very safe in case something goes wrong, like a short-circuit on a connector or perhaps a power surge going through).

Cosindering the price to refurbish this and the extra electricity it takes, some new half-decent 350watter, 80 PLUS (Bronze) rated would pay itself in a year, most likely.

Sure.
But there is also the question of how much the new PSU will last. With crap caps, it could fail again shortly after the warranty, and very likely after 5 years. Meanwhile, these old Deer/Allied/L&C cheap PSUs are so crude that there is almost nothing to fail in them. It's both sad and funny, really. However, with good caps and a few small component upgrades, you get a working PSU that is likely to outlast many low-end decent PSUs on the market. So all in all, it's a trade-off. Yes, you pay a tiny bit more for electricity, but you also have a really simple PSU that is less likely to fail on you... though on that note, I guess it all depends on how long the fan lasts. If you recap a PSU and don't take time to make sure the fan is in good working order, then that would be the next failure point.

Bad caps matter, but he would get crap caps anyway if he bought a new psu for 20-30€. Try convincing people here to spend more for a 2012 PC. He was thinking of recycling it!

Exactly!

It appears that most "common" people seem to think that when something breaks, it can never be as good as a new item, even if fixed properly. Thus, many folks often don't like to "invest" money into a broken PC, even if I try to convince them that the repaired one is likely to outlast whatever new PC they can buy on the market now. So these are the people that want the fix done as cheap as possible. And in the end, all they look at is the bottom line of your repair. Doesn't matter if you tell them that you're putting this slightly more expensive but much better PSU that will save them more money in electricity in the long run. They would rather save some money NOW (from the repair) than be given the promise of saving more money LATER (from less electricity usage).

Have I posted these ones?
Controller seems to be a PIC - so some kind of Digitally controlled thing...

Have no idea how that thing works...
And it seems to regulate every voltage independently with its own transformer?!

I don't remember if you posted it or not, but I do recall seeing something similar a very long time ago.

It's a very strange design, indeed. Looks almost like each rail is mag-amp regulated, or something like that. But who knows. In any case, I would NOT want to try to troubleshoot that if something went wrong with it. Granted, at least I see lots of through-hole resistors there, so looks like there may not be that many SMD parts (if any?). That will certainly make troubleshooting easier, but not by much, simply due to the complexity of this PSU.

[Behemot]

Re: Power supply build quality pictorial. part 2

I myself prefer to refurbish units from OEM machines whenever possible (like Delta, HiPro, LiteON, and Bestec from Dell/HP/Gateway/Acer PCs.) But I wouldn't call these more "robust". I think a better way to describe them is to say they are safer (whereas the really cheap PSUs could be just as robust... with good caps and fan, of course..., but not very safe in case something goes wrong, like a short-circuit on a connector or perhaps a power surge going through).

Also not laughable heatsinks, higher rated silicon and all that stuff.

Sure.
But there is also the question of how much the new PSU will last. With crap caps, it could fail again shortly after the warranty, and very likely after 5 years. Meanwhile, these old Deer/Allied/L&C cheap PSUs are so crude that there is almost nothing to fail in them. It's both sad and funny, really. However, with good caps and a few small component upgrades, you get a working PSU that is likely to outlast many low-end decent PSUs on the market. So all in all, it's a trade-off. Yes, you pay a tiny bit more for electricity, but you also have a really simple PSU that is less likely to fail on you... though on that note, I guess it all depends on how long the fan lasts. If you recap a PSU and don't take time to make sure the fan is in good working order, then that would be the next failure point.

Have not really seen anything else than FSP and very few Seasonic OEM units with bad caps older than 8 years lately. The caps will eventually likely fail, but with higher efficiency and most likely also some tiny progress from cap manufacturers, this time is constantly getting longer. So long most ppl will throw these cheap units away long before the PSU fails on caps.

More often something of the protection/feedback components goes bad. Also some CWT (GP) platforms seem to fail because of components around SB controller exploding

Magamp regulation was used for older server/enterprise units for both the +3.3 and +5V rails, you can easily keep almost rock-stable regulation while only driving the single rail - +12 V. SAMA seems to use that for some of their lower-mainstream platforms. The next step seemed to be DC-DC and digital controllers, todays server PSUs are so complicated I often barely get heads and tails from it.